In opportunistic networks, nodes communicate intermittently based on storecarry-forward paradigm while exploiting node mobility. The challenge is to determine the ideal nodes to deliver the messages since there is no end-toend connectivity. The nodes might make this decision based on the data sensed from the network. This technique is not ideal in scenarios where the speed of changes in the network topology is greater than the speed at which the nodes can collect info on the network, which might, in turn, be restricted due to usage constraints and uncertainty of knowledge about future contacts. To tackle the problems raised by the non-deterministic environments, in this paper, a stochastic optimization model and corresponding algorithm are developed to find the optimal routes by considering the short and long-term impact of choices, ie, the next hop. Herein, we first propose a stochastic model to resolve the routing problem by identifying the shortest path. In the second step, we show that the optimal solution of the proposed model can be determined in polynomial time. An online algorithm is then proposed and analyzed. The algorithm is O(lognρ) competitive considering the number of nodes and their associated energy. This model can take advantage of the unexpected meets to make the routing more elastic in a short time of contact and with less of a burden on the buffer. The simulation results, against the prominent algorithms, demonstrate significant improvement of the proposed approach in delivery and average delay ratio.KEYWORDS opportunistic networks, stochastic optimization, online routing, competitive ratio
| INTRODUCTIONOpportunistic networks (ONs) are the kind of networks where nodes are wirelessly and erratically connected using an ad hoc like structure. The communications between the nodes happen without a fixed communication infrastructure to support the mobility needs of the nodes. The challenges to be tackled by these networks include sporadic connection, long and variable delay/errors, and the fact that no end-to-end connectivity exists for these sparse and self-organized networks. When a node wants to transmit a message to another node with no direct connection, packets can be forwarded using relaying nodes to facilitate the packet delivery from the originating node to the final destination. Inherently, each node supports the store-carry-forward paradigm within the network. 1 Accidentally, moving nodes can facilitate intermittent connectivity cycle towards the ultimate destination. In some literature, the terms ON and delaytolerant network (DTN) are often used interchangeably. In our view, ONs are indeed a subset of DTNs (DTNs assume